Today was my last day @climateai. I have had the opportunity to work with so many amazing people over the last two years. Coming straight out of academia and being thrown into an early-stage startup made for an insane learning experience. Here are some take-aways:
1) Start-ups are chaotic but provide a super-charged learning/growth environment. Less deep on science but so much more on business, management, hiring, product development, data and software engineering, etc. Requires a proactive disposition to thrive though.
2) Talking about data engineering: It's the main bottleneck for companies in the climate/EO space. Huge data volumes on the cloud coming from different sources need to be updated frequently and reliably. Very few people with full stack expertise out there, so development is slow.
3) The @pangeo_data stack (xarray, dask, zarr, ...) is incredibly useful and should be part of basic education in wx/climate degrees. But it's not enough for making things operational. Workflow management (e.g. Airflow), REST APIs and CI/CD are other crucial ingredients.
4) In all companies there is a tension between business/sales and science. They inhabit different worlds with different standards but they depend on each other. It requires non-stop work to find common ground. Often, that requires letting go of some idealism on the science side.
5) In academia, forecast skill reigns supreme. In the real world, wx/climate forecasts are only one component in very messy decision making processes. Having impact isn't as simple as building a better model. We (researchers) should put more effort in talking to end users.
